Causal inference techniques are now more and more applied in all fields of epidemiological, medical, political and social science where ideally randomized trials are not available but, e.g., observational studies. Different approaches -- at a first glance seemingly unconnected - have been developed and make the field somehow mystique. An important reason is that the concept of causality is difficult to treat within the theory of probability and distributions without further formal support.
Therefore a workshop on casual inference was organized by Leonardo Briceño Ayala, lecturer and long-term partner of CIHLMU , at the Universidad del Rosario in Bogotá, Colombia in cooperation with Prof. Dr. Christian Heumann, LMU lecturer in Methods of missing data, Model selection and Model averaging. The main learning objectives of the course were to give a step-by-step introduction to the field and to enable the participants to understand the new developments and incorporate them in a critical manner into their own studies.
The three-day course covered the following contents:
- Introduction to causal inference
- The Potential Outcomes Approach (POA) or Counterfactual Outcomes
- Introduction to directed acyclic graphs (DAG) and causal DAGs.
- Inverse probability weighting
- Structural mean models
- G-formula and G-Estimation
- Balancing methods (Propensity score, matching)
- Instrumental variables
- Targeted Maximum Likelihood
- Critical view on the POA approach
After successful completion of the workshop including group work, the ten participants from different Latin American countries received a certificate of 3 ECTS. CIHLMU scholarships for partners and alumni also enabled participants from low and middle income countries to participate.